ObjectiveTo explore the potential mechanisms of action of the modified Duhuo Jisheng Mixture （JDJM） in treating synovial lesions in knee osteoarthritis （KOA）. MethodsA total of 43 male New Zealand white rabbits were randomly allocated into a blank group （n=8） and a model group （n=35）. The KOA model was induced by immobilizing the right hind limb with a high-molecular resin plaster bandage， with a modeling period of 6 weeks， resulting in successful modeling in 32 rabbits. These rabbits were then randomly allocated to the model group， celecoxib group， JDJM group and JDJM+740Y-P group， each consisting of 8 rabbits. The celecoxib group received celecoxib via gavage at a single dose of 0.009 3 g·kg^-1， while the JDJM was administered a single dose of 6.8 mL·kg^-1 （4.515 2 g·kg^-1） of the herbal preparation via gavage. The phosphatidylinositol 3-kinase/protein kinase B/mammalian target of rapamycin （PI3K/Akt/mTOR） pathway activator + JDJM group received 4.515 2 g·kg^-1 of the herbal preparation via gavage along with an auricular vein injection of 0.15 μmol·kg^-1 740Y-P. For a period of 6 weeks， the remaining groups received an equal volume of physiological saline via gavage daily. After the medication period， the knee joint pain threshold and circumference were measured， and hematoxylin-eosin （HE） staining was performed to assess the pathological changes in the synovial tissues. Enzyme-linked immunosorbent assay （ELISA） measured the levels of interleukin-1β （IL-1β）， interleukin-6 （IL-18） and tumor necrosis factor-α （TNF-α） in the joint fluid. Real-time fluorescence quantitative polymerase chain reaction （Real-time PCR） was used to assess the mRNA expression of PI3K， Akt， mTOR， NOD-like receptor protein 3 （NLRP3）， cysteine-requiring aspartate protease-1 （Caspase-1） and gasdermin D （GSDMD） in the synovial tissues. Immunohistochemical （IHC） assay was performed to assess the protein expression of NLRP3， Caspase-1 and GSDMD. Western blot was carried out to analyze the protein expression of p-PI3K/PI3K， p-Akt/Akt， p-mTOR/mTOR， NLRP3， Caspase-1 and GSDMD. ResultsCompared to the blank group， the model group showed a significant increase in knee joint circumference and decrease in pain threshold， the synovial tissue pathology score was higher （P<0.05）， and the levels of IL-1β， IL-18， and TNF-α in the joint fluid significantly increased （P<0.01）. PI3K， Akt， mTOR phosphorylation as well as mRNA and protein expression increased （P<0.01）， while the mRNA and protein expression levels of NLRP3， Caspase-1 and GSDMD also significantly increased （P<0.01）. Compared to the model group， the celecoxib and JDJM groups exhibited a significant reduction in knee joint circumference and increase in pain threshold， the synovial tissue pathology score was lower （P<0.05）， and the levels of IL-1β， IL-18， and TNF-α in the joint fluid decreased （P<0.01）. The mRNA and protein expression of p-PI3K， p-Akt， p-mTOR， NLRP3， Caspase-1 and GSDMD were reduced （P<0.01）. Compared to the JDJM group， the JDJM+740Y-P group showed a decrease in the improvement of synovial lesions， an increase in knee joint circumference， and a decrease in pain threshold. The synovial tissue pathology score was lower （P<0.05）， and the levels of IL-1β， IL-18， and TNF-α in the joint fluid were higher （P<0.01）. The mRNA and protein expression of p-PI3K/PI3K， p-Akt/Akt， p-mTOR/mTOR， NLRP3， Caspase-1 and GSDMD increased （P<0.01）. ConclusionJDJM is effective in treating KOA. Its mechanism may involve modulating the PI3K/Akt/mTOR pathway in synovial tissues， inhibiting pyroptosis， reducing inflammatory factor release， and protecting bony structures.

We proposed a new deep learning model by analyzing electroencephalogram signals to reduce the complexity of feature extraction and improve the accuracy of recognition of fatigue status of pilots. For one thing, we applied wavelet packet transform to decompose electroencephalogram signals of pilots to extract the δ wave (0.4-3 Hz), θ wave (4-7 Hz), α wave (8-13 Hz) and β wave (14-30 Hz), and the combination of them was used as de-nosing electroencephalogram signals. For another, we proposed a deep contractive auto-encoding network-Softmax model for identifying pilots' fatigue status. Its recognition results were also compared with other models. The experimental results showed that the proposed deep learning model had a nice recognition, and the accuracy of recognition was up to 91.67%. Therefore, recognition of fatigue status of pilots based on deep contractive auto-encoding network is of great significance.